The Discovery Conversation

Key Concepts

First Principles Thinking - Instead of asking "how do I make this process better?", ask "what is actually needed here, and what's the simplest path to it?" This means rejecting the framing you're given and working backward from the true underlying need.

Symptom vs. Problem - The issue someone describes is almost always a symptom. "The report takes two days" is a symptom. "Leadership lacks current competitive data when making decisions" is the problem. Building for the symptom produces something that technically works but doesn't help.

Two Doors of Discovery - Discovery either starts with an existing workflow you need to understand deeply enough to improve, or a gap where no process exists and you need to identify what to build from nothing. Both require the same core skill but different approaches.

Quick Reference

Mapping an Existing Workflow

Before touching the process itself, define the boundaries:

• What does "done" look like for this person?
• What systems and tools are involved? Where does data live?
• How is the work measured? What makes it good vs. bad?
• Where does the stakeholder think automation fits? (This reveals both opportunity and constraints.)

Five Techniques for Extracting the Real Process

1. Anchor to a specific instance. Don't ask "how do you do this?" Ask "take me back to the last time you did this. What did you open first?" Memory surfaces mess. Descriptions surface summaries.
2. Stay quiet for two full minutes. After asking, don't interrupt or follow up. People fill silence with the steps they'd otherwise skip, especially the ones they dismiss as unimportant.
3. Decompose every verb. Words like "review," "compare," and "clean up" hide multiple operations. When you hear one, ask: "What does review actually look like?"
4. Probe the three universal gaps:
   • Before the work starts - "What's the very first thing you open?" Catches hidden inputs.
   • After the work is done - "After you send it, is there anything else?" Catches cleanup steps that enable the next run.
   • When things go wrong - "Has this ever gone sideways?" Every failure story is an unwritten requirement.
5. Run two passes. First pass: breadth. Get the full shape without digging in. Second pass: depth. Return to each chunk and live in it. People cannot do both simultaneously.

Discovering What's Never Been Built

Look for three things across the organization:

• Data nobody's using - signals ignored because of volume or format
• Oversight nobody's doing - things that should be monitored but aren't
• Things people assume are impossible - requests that never make it to a wish list

Adoption Reality Check

• Enthusiasm in discovery does not predict adoption. People overestimate their excitement and underestimate the friction of behavior change.
• Whatever you build must be 10x better for them and twice as easy as the current approach.
• Think horizontally across teams. The problem one team describes is often a category of problems shared by several.

Pre-Conversation Prep

1. Write down what you think the problem is
2. Write down three ways you might be wrong
3. During the conversation, if you catch yourself thinking about what to build, force yourself back to what is actually happening
4. Default to asking why over how - "why do you do it this way?" reveals logic that "how do you do this?" misses